bigrquerystorage

Comparing bq_table_download from bigrquery to bgs_table_download from bigrquerystorage

The main utility is to replace bigrquery::bq_table_download method.

It supports BigQueryRead interface. Support for BigQueryWrite interface may be added in a future release.

Advantages over BigQuery REST API

BigQuery Storage API is not rate limited and per project quota do not apply. It is an rpc protocol and provides faster downloads for big results sets.

Details

This implementation use a C++ generated client combined with the arrow R package to transform the raw stream into an R object.

bqs_table_download is the main function of this package. Other functions are helpers to facilitate authentication and debugging.

Installation

CRAN

install.packages("bigrquerystorage")

Github

remotes::install_github("meztez/bigrquerystorage")

System requirements:

gRPC
protoc

Debian 11 & 12 / Ubuntu 22.04

# install protoc and grpc
apt-get install -y libgrpc++-dev libprotobuf-dev protobuf-compiler-grpc \
                   pkg-config

Fedora 36 & 37 & 38 / Rocky Linux 9

# install grpc, protoc is automatically installed
dnf install -y grpc-devel pkgconf

Other Linux distributions

Please let us know if these instructions do not work any more.

Alpine Linux

apk add grpc-dev protobuf-dev re2-dev c-ares-dev

Alpine Linux 3.19 and Edge do not work currently, because the installation of the arrow package fails.

Debian 10

Needs the buster-backports repository.

echo "deb https://deb.debian.org/debian buster-backports main" >> \
    /etc/apt/sources.list.d/backports.list && \
    apt-get update && \
apt-get install -y 'libgrpc\+\+-dev'/buster-backports \
    protobuf-compiler-grpc/buster-backports \
    libprotobuf-dev/buster-backports \
    protobuf-compiler/buster-backports pkg-config

OpenSUSE

In OpenSUSE 15.4 and 15.5 the version of the grpc package is tool old, so installation fails. You can potentially compile a newer version of grpc from source.

Ubuntu 20.04

In Ubuntu 20.04 the version of the grpc package is tool old, so installation fails. You can potentially compile a newer version of grpc from source.

CentOS 7 & 8 / RHEL 7 & 8

These distros do not have a grpc package. You can potentially compile grpc from source.

macOS

If you use Homebrew you may install the grpc package, plus pkg-config. If you don’t have Homebrew installed, the package will download static builds of the system dependencies during installation. This works with macOS Big Sur, or later, on Intel and Arm64 machines.

brew install grpc pkg-config

Windows

From Rtools43, grpc is included in the toolchain.

The package used to automatically download static builds of the system requirements during installation but this was removed per CRAN policy. Only, R 4.3.x (with Rtools43) or later is currently supported.

Example

This is a basic example which shows you how to solve a common problem. BigQuery Storage API requires a billing project.

# Auth is done automagically using Application Default Credentials.
# or reusing bigrquery auth.

# Use the following command once to set it up :
# gcloud auth application-default login --billing-project={project}

library(bigrquery)
library(bigrquerystorage)

# TODO: (developer): Set the project_id variable to your billing project.
# The read session will bill this project. This project can be
# different from the one that contains the table.
project_id <- 'your-project-id'

rows <- bqs_table_download(
  x = "bigquery-public-data:usa_names.usa_1910_current",
  parent = project_id
  # , snapshot_time = Sys.time() # a POSIXct time
  , selected_fields = c("name", "number", "state"),
  row_restriction = 'state = "WA"'
  # , sample_percentage = 50
)

sprintf(
  "Got %d unique names in states: %s",
  length(unique(rows$name)),
  paste(unique(rows$state), collapse = " ")
)

Authentication

Done using Google Application Default Credentials (ADC) or by recycling bigrquery authentication. Auth will be done automatically the first time a request is made.

bqs_auth()
bqs_deauth()

Stability

Does not support AVRO output format. Report any issues to the project issue tracker.

Full gRPC debug trace with bigrquerystorage:::bqs_set_log_verbosity(0).